A novel drilling parameter optimization method based on big data of drilling

Rate of penetration (ROP) is the key factor affecting the drilling cycle and cost, and it directly reflects the drilling efficiency. With the increasingly complex field data, the original drilling parameter optimization method can't meet the needs of drilling parameter optimization in the era of big data and artificial intelligence. This paper presents a drilling parameter optimization method based on big data of drilling, which takes machine learning algorithms as a tool. First, field data is pre-processed according to the characteristics of big data of drilling. Then a formation clustering model based on unsupervised learning is established, which takes sonic logging, gamma logging, and density logging data as input. Formation clusters with similar stratum characteristics are decided. Aiming at improving ROP, the formation clusters are input into the ROP model, and the mechanical parameters (weight on bit, revolution per minute) and hydraulic parameters (standpipe pressure, flow rate) are optimized. Taking the Southern Margin block of Xinjiang as an example, the MAPE of prediction of ROP after clustering is decreased from 18.72% to 10.56%. The results of this paper provide a new method to improve drilling efficiency based on big data of drilling.